The ATP-dependent chromatin-remodeling complexes play important roles in gene regulation by opening chromatin structures for transcriptional activators or repressors. The prototype of this type of complexes is the SWI/SNF complex, which was found in diverse organisms, including yeast, Drosophila, mouse and human. It is required for proper expression of homeotic genes and segmentation in Drosophila, and mutation in one subunit of the complex causes pediatric rhabdoid cancer in humans. We have purified several human SWI/SNF-related complexes. By microsequencing, we have identified and cloned most of the subunits from the major form of the complex. We have microsequenced and cloned the largest subunit, BAF250. Sequence analysis revealed that BAF250 contains a DNA binding domain similar to yeast SWI1, and several LXXLL motifs, which have been previously shown to be able to interact with nuclear hormone receptors. Using transient transfection assays, we found that BAF250 in fact facilitates transcriptional activation by glucocorticoid receptor (GR). The region containing LXXLL motifs of BAF250 also interacts with GR in vitro. This work suggests that BAF250 may be a targeting subunit of hSWI/SNF, and may mediate the recruitment of the complex to DNA-bound glucocorticoid receptors. Much of this work has already been published in Mol. Cell. Biol. As a continuation of this project, we have cloned a novel human homolog of BAF250a, termed BAF250b. The two genes share over 60% of identity and possess same type of domain structure. We have now isolated a BAF250b-containing complex. It shares several identical subunits with BAF250a complex but also contains its own unique components. One unique subunit is ENL, a fusion partner for MLL which is a common target for chromosomal translocation in human acute leukemia. ENL is also a human ortholog of yeast SWI/SNF subunit, TFG3. We demonstrate that the resultant MLL-ENL fusion protein also associates with a human SWI/SNF complex. Moreover, the fusion protein cooperates with SWI/SNF to activate the promoter of HOXA7, which is a downstream target of MLL and essential for oncogenic activity of the MLL fusion proteins. Our data suggest that human SWI/SNF complexes show considerable heterogeneity, and one or more may be involved in the etiology of leukemia by functioning with MLL-fusion proteins. This work has been published recently in MCB. We are continuing to identify the genes that are specifically dependent on BAF, but not PBAF, for expression. Using siRNA, we were able to deplete the BAF-specific subunit, BAF250. We showed that one interferon-responsive gene, IFGM3, specifically depends on BAF but not PBAF for expression. This result demonstrates that BAF and PBAF have selectivity in mediating expression of different genes. It was included in publication in Genes & Development. We plan to investigate the mechanism of how BAF is targeted to IFGM3 and other genes.